Power amplification, or the last stage of gain in a hi-fi system, can be described in its simplest form as a ‘modulated power supply’ required to drive a loudspeaker.
Historically, they can be categorised into three main groups: class A solid state and tube; Class AB solid state and tube; then variants, e.g. Class D, etc.
The prime function of a power amp, like the raison detre of all good hi-fi, is to add nothing nor to take anything away from the signal being amplified.
In the 1960’s when transistors first became widely useable in domestic audio applications, class AB became the norm where the original signal is shared between two opposite polarity devices in a push-pull design. This was and is, an efficient method of providing good levels of gain for a moderate investment in devices including the power supply. Push-pull was also the norm for valve amplifiers in the later years of their rule, but unlike transistors, where there is an npn-pnp device available, in order to achieve the necessary negative and positive halves of the push-pull signal, a phase splitting device had to be employed.
However, the methods usually employed were never accurate enough to perform the operation of two exact mirror images of the signal. In addition, to match the demands of ever more current hungry and falling impedance loudspeakers, more devices were employed which gave the necessary power at the expense of ‘switching distortion’ due to the class B operation of the amplifying devices, which although almost impossible to measure, is very obvious to the ear.
Single-ended class A tube amplification, the true foundation of all hi-fi amplifiers, have existed since the 1920’s, but inevitably have suffered to date from their own equally as bad list of problems.
Class A, single-ended amplifiers were and are the ‘holy grail’ of many enthusiasts, but suffer from almost masochistic, hair shirt like problems. Low outputs are the norm, requiring super efficient speakers. A valve amp of this type has typically a liquid euphonic mid range with a greatly reduced bandwidth creating totally unrealistic reproduction at frequency extremes. Levels of distortion which would be unacceptable if found in solid state amplification are tolerated as in some cases they can have a cancelling effect to some of the distortion added by the loudspeaker itself. Some valves have a notoriously short life but all need to be biased properly to perform at their best.
Class AB valve amps fall into the same trap as their solid-state cousins, by splitting the signal and in addition, forcing the tube to exist in a far from ideal environment. This creates added noise and distortion.
All traditional valve amplifiers suffer from the fact that their control circuitry may have changed in terms of the components utilised, but have changed little in concept and design in the last 40 years.
There is however more to the problem than that. In order to create a believable stereo image (‘stereo’ from the Greek ‘stereos’, meaning solid or combining form) many critical problems are required to be solved.
Why the need for ultra wide bandwidth?
To ensure that phase is kept intact throughout the audio bandwidth, it is of no use to build a device with a 20Hz – 20KHz bandwidth (the theoretical limit of human hearing) as phase shift will destroy the subtle spatial cues contained in the signal and not only strip the harmonics from the fundamentals in the music, but in short cause the sound arriving at the listener to bear little resemblance to the sound arriving at the microphone on the day of the recording. The result is a distorted image of the performance with tonal aberrations. It will never get even close to reproducing the original performance.
So why use valves? First and foremost, valves are inherently very linear; secondly, a well-designed valve amplifier can swing more (class A) voltage at its output than most solid-state designs, providing a far more realistic portrayal of dynamics. Additionally by using a properly designed output transformer, the back EMF (the voltage generated by the speaker coil moving within its magnetic field and flowing back to the amplifier), which in solid-state designs cause distortion, is much reduced. most Solid-state amplification suffers from a percentage of higher order harmonic distortion ie, 3rd,7th,9th, etc, which even in small amounts is widely accepted to be much more objectionable than the often higher amounts of lower order harmonics ie, 2nd, 3rd produced by valve amplification, therefore’ in most cases the distortion produced by a valve amplifier is of a more audibly acceptable type.
O.t.l (output transformerless) valve amps suffer from the same disastrous failings as solid state designs! also valves are high impedance devices and therefore not intended to drive low impedance loads ie; loudspeakers direct!
By choosing as a goal to design an ultra low noise, ultra wide bandwidth, hyper linear amplifier range, and by utilising the best attributes of two technologies (tube and solid state) it was decided three years ago to rip up the book that had its grounding in the 1930s, and devise a means of fulfilling the promise and capability of valves that had never even been approached with conventional circuitry.
Anthony Matthews of Tube Distinctions began researching the project which was to take him backwards in time to look at designs From 60 years ago, and forward in time to discover the fundamental physics that governed the flow of electrons within a vacuum valve/tube.
He deduced that most valve designs suffer from the same problems i.e. they rely on one valve to drive another, where in common practice the first valve has an impedance so high that it is incapable of driving much at all.
The parasitic (i.e. the support circuitry and hardware) directly affect the performance as the capacitance of the board, the cable and poor, archaic circuit topology destroy the top end bandwidth; combine this with a high output impedance of the first valve and you have a situation akin to a ‘motor cycle towing a truck’. The transient response by this time has died a death; plenty of mid range but no extremes and no control.
To put this in perspective, let us look at what mainstream market leaders of valve amplification publish about their products.
Distortion is typically measured at 1 watt. It is all too obvious why, in that any more power produces more distortion. If we were happy to listen at 1 watt, all would be well, but it is obviously unrealistic for most listeners.
• Distortion figures are published measured at 1 KHz, giving a tiny snapshot of the true picture. If they came clean with the real figures across the whole bandwidth, chances are it would be a completely different story.
• Accurate noise figures of valve amps are rarely published, as rather like a lady of a certain age, they are very conscious of their size.
• Distortion figures of 0.5 and even 1%: This sounds impressive but compared with solid-state amplification of a similar size is hundreds of times greater.
• Bandwidth figures measured at again 1 watt: in order to give the best impression those brave enough to publish full power bandwidth figures attempt to fudge the issue quoting –3dB points, showing conclusively that they are not 24 Bit compliant.
• Phase response is rarely mentioned.
The previous info is derived from the current promotional material of the market leaders in valve amplification. What makes this all the more meaningful is that it relates to class AB amplification. The statistics from our direct competitor i.e. single ended Class A amplification shows typically bandwidths of 20Hz – 25 KHz with noise and distortion figures of ‘Less than 3%’.
Due to the limits of conventional amplifier technology widely used up to now, we have somehow come to regard these statistics as impressive and even virtuous.
It is strange to relate that most conventional single ended valve amps are unable to reproduce the bandwidth of most high-end cartridges, let alone the dizzy heights achievable by DVDA !!!
We feel that a change is long overdue.
In the Soul Series amplifiers, the valve is placed, arguably for the first time in hi-fi history, in a totally controlled environment.
In conventional designs, the current for the valve (i.e. the current that determines and controls how much power can be drawn by the valve) is set usually by a capacitor and a resistor. This is completely inadequate, as the goalposts are constantly moving with the signal. Putting the valve into a stressed and fluctuating environment where it can never reach a steady state. This additionally reduces its short life.
To control the valve perfectly, seven circuits were designed with approximately 200 components controlling each single valve 15-watt power module. These lock the valve into the ideal linear operating environment. Each of the circuits can be run independently of each other, most of which are populated with surface mounted devices and our own discrete hyper-regulators, with proprietary voltage references.
This provides a noise floor at the output of the amplifier in the microvolt domain. With a Soul Series on full gain, very little noise can be heard at the speaker, even with one’s ear up against the cone! Conventional valve amplifiers have a noise floor typically of around 50 mv or more, some 50 times greater, burying dynamic range and resolution and certainly audible at full gain without one’s ear to the cone! This unique approach has other benefits, as it transforms the reliability and life of the valve. We are quoting figures of a three year useful life, but we predict that many will last longer.
Why push the limits of the bandwidth and why is it important?
As discussed earlier, the wider the bandwidth that can be obtained, the more linear the signal circuit design and the lower the phase errors in the audio band width.
This means that faithful three-dimensional sound staging and imagery, with true perspective and harmonic realism, will for the first time be obtained from a valve with the added benefit of the lowest distortion ever from a class A, single-ended design. This unique approach reproduces the recording in way never before achievable, as most of the negative effects of amplification we have come to unconsciously accept are no longer present.
In the worse case scenario, our smallest version of the Soul Series, where a single valve 15watt module is driving a loudspeaker via it’s output transformer, it still produces a ‘flat’ frequency bandwidth of 20Hz to over 120KHz. In addition, the phase remains flat to 60 KHz.
To obtain more power the modules are paralleled. This has a threefold effect…
• For every paralleled pair the power doubles.
• For every paralleled pair the noise is reduced.
• For every paralleled pair the distortion is reduced.
The measured performance of the 30-watt stereo has an increased bandwidth flat from 20Hz to over 130 KHz, with a phase response flat in excess of 60 K.
This means that the Soul Series has phase accuracy 5 times greater than the signal bandwidth of most of the single ended amps ever built.
Unlike conventional amps, the signal circuitry in the Soul Series is semi-floating and not directly referenced to mains earth. This is vital as mains born pollution can add to noise on the signal, which reduces resolution and dynamic range.
Unlike other single-ended amps, the Soul Series clips symmetrically; those of you who have heard a conventional single-ended triode amp clip will know what we mean.
All signal circuitry is DC coupled (no servo’s), compared with conventional valve technology where AC coupling via capacitors is the rule. Many of our competitors try to improve this by extolling the virtues of the use of Polypropylene and Teflon variants. All types of capacitors need time to charge and discharge. The more expensive Polypropylene and Teflon caps are faster than standard types but still add time delays – phase shift. The use of any capacitor in the signal path will add coloration, reduce resolution and make the amplifier react slower to transients, which in addition destroys phase integrity of the signal, which will produce not only a distorted sound stage in terms of space, but the vital harmonic structure of the sound will be shifted in time. The bigger the value of the cap, the more pronounced the problem becomes. This is particularly noticeable in the bass, where bloom and overhang mask the true harmonic structure and speed of the notes. This is partly the reason why in extreme complicated musical passages the integrity of most of these amplifiers’ falls to bits.
The extreme linear/ noise properties of the Soul Series amplifiers let you play the music you want, not what your system dictates!
It has taken a little over three years, but we feel we have created arguably the ultimate in valve/semiconductor amplification. Ready to tackle even the most demanding of frequency extremes – to date, 24bit 96k!
A comment that has been levelled at our other products has been made to describe the sound of this new amplifier…“I cannot pin it down”.
The truth is, because the Soul Series has little or no characteristic traits of its own, for arguably the first time ever you are able to hear in much more detail what’s actually there.
Three models are available, from the modest but very capable 15-watt stereo, to the stunning 60-watt monoblocs.
• 15 watt stereo – Price on application.
• 30 watt mono-block – Price on application.
• 60 watt mono-block – Price on application.
” All men dream, but not all equally. Those who dream by night, in the dusty recesses of their mind, wake to find it was all vanity. But the dreamers of the day are dangerous, for they may act their dreams with open eyes to make things happen.”T.E. Lawrence, (Lawrence of Arabia)